On June 3, 2026, Ray Dalio went on television to explain how the AI bubble ends. Not with a technological failure, he said, but with a conversion. A bubble bursts when holders of paper wealth are forced to turn it into spendable money — when the tax bill, the margin call, or the redemption arrives, and a billion-dollar valuation has to become dollars that can actually be spent. Investors, he warned, confuse wealth with money. They are not the same thing. Wealth is what you own. Money is what you can spend right now.
That distinction is older than any bubble, and it is the quiet hinge of this essay. For all of history, what you own and what you can spend have been two different things, separated by an act of conversion that was slow, costly, and occasionally — at the worst possible moment — impossible. We bridged the gap by holding money: keeping part of our wealth in the spendable form at all times, as insurance, and paying for the privilege in the yield we gave up to hold it.
That bridge is now becoming nearly free to cross.
A companion essay argued that the dollar’s deepest power is settlement power, and that its grip on how the world pays is loosening. This one looks at a stranger possibility one level down: not which money the world uses, but whether we will go on holding a dedicated spending money at all.
The cost of exchange is collapsing
Start with what can be asserted plainly, because it is happening on the record.
The conversion of one form of value into another is getting radically cheaper, and the assets you can convert are multiplying. On June 1, 2026, Binance — the largest retail exchange on earth — opened more than 7,000 U.S. stocks to non-U.S. users, bought with stablecoins, and previewed “bStocks,” tokenized shares that live on a blockchain. It is not the first: tokenized U.S. equities have traded on-chain since 2025, with Kraken’s xStocks crossing tens of billions in volume. And the largest asset manager in the world is pushing the same direction: in his 2026 letter, BlackRock’s Larry Fink compared tokenization to the internet in 1996 and described a single wallet holding every asset a person owns — stocks, bonds, funds, currencies — side by side. On-chain real-world assets passed $30 billion in 2026, roughly tripling in a year, with one consultancy projecting $16 trillion by 2030.
Two things follow, and only the first is certain. The certain one: access is no longer scarce. A saver in Lagos or Jakarta can hold a fraction of an American company, settled in seconds, for cents — something that a few years ago required a domestic brokerage account most of the world could not open. The system is still being walled with the old tools — KYC, geofencing, whitelisted wallets, and a 2026 regulatory interpretation insisting that tokenization does not relieve sanctions screening — and, tellingly, U.S. retail is still mostly fenced out while the rest of the world is let in. But a token in self-custody is harder to gate than an account in a correspondent bank. The wall still stands; it has simply become porous.
So here is the first question, and it is not rhetorical: when any asset can be turned into any other at the moment of payment, for almost nothing, what is left that only money can do?
What “free” used to cost
It helps to remember that the access we already had was never actually free.
The companion to this essay traces the hidden price in detail — the yield a bank keeps on your deposit, the float earned during the days a payment clears. The same was true the last time finance “democratized.” Retail trading went commission-free a decade ago, and the bill did not vanish; it was moved out of sight. Brokers sold their customers’ orders to market makers, who paid for the right to be on the other side — an arrangement that earned the industry $4.8 billion in 2025 while the trades read as free. The cost was real. It was simply unpriced, paid in worse execution and in the value of letting someone see what you were about to do.
That is the before-picture. The thing on-chain settlement does is not invent cheap exchange; it makes the price of exchange a quoted, visible number — and a price that can be seen can be competed toward zero. The question that opens is what remains once it is.
Money is the tax we paid for liquidity
Of money’s three textbook jobs — store of value, unit of account, medium of exchange — only the last exists purely because conversion was expensive.
You do not hold cash because you want to. Cash pays little and loses to inflation. You hold it because you will need to transact, and turning your real wealth — your home, your shares, your claim on a business — into spendable form was, until now, slow and costly and uncertain. The held medium of exchange is insurance against that friction: a buffer of pre-converted wealth you keep on hand so you are never caught needing to spend what you cannot quickly sell. The yield you forgo to hold it is the premium on that insurance.
This is the human fact underneath the whole story, and it runs the other way from how we usually tell it. People have never wanted to hold money. Across every era they have tried to keep their wealth in the best store of value they could find — land, gold, livestock, equity — and held only as much spending money as the friction of the age forced on them. The medium of exchange was a concession to transaction cost, not a preference.
So the question sharpens. If the friction goes to near zero — if any asset can become spendable at the instant you spend it — does the concession still have to be made? Do you keep paying the premium on insurance against a risk that is disappearing?
The machine that converts for you
The reason this is no longer idle is that the converting is starting to be done by software, not by you.
A program can now hold a balance, watch a price, and settle a payment without a human moving through a bank’s interface. Early machine-payment rails are already live — one protocol crossed 100 million agent-initiated transactions on a single network within nine months — and beneath them sit “intent” systems where you state what you want and competing solvers find the cheapest route automatically. The transaction counts are real even where the dollar amounts are still small and much of the early traffic is experimental; the machinery exists, and it is built to drive conversion toward its marginal cost.
Follow that one step. If an agent can turn your best-held asset into the exact spendable form a payment requires, at the moment it is required, then there is no longer an obvious reason to pre-hold a buffer of idle, low-yielding cash. You would hold the thing that pays you, and let the machine mint the medium of exchange just in time, for the instant it is needed, and let it dissolve again.
Which leaves the strangest question of all: in such a world, what are you actually holding between transactions — and is “money,” as a thing you keep, even the right name for whatever fleeting, just-in-time form the spending takes?
The three functions come apart
If holding the medium of exchange becomes optional, money’s three jobs, fused for millennia because conversion was costly, can finally separate — and the early evidence is that people are already pulling them apart by hand, before any agent does it for them.
Americans now hold a record $7.89 trillion in money-market funds — yield-bearing near-cash they hold instead of a checking balance. Tokenized Treasuries, the on-chain version of the same instinct, grew past $15 billion from almost nothing in two years. The honest reading is that much of this is simply a high-rates story — money fleeing zero-yield deposits because rates are high. But the behavior is exactly the one the technology is built to accelerate: refusing to hold the medium of exchange when holding it costs you.
Push it forward and the functions land in different places. The unit of account — the ruler we quote prices in — is sticky, and likely stays the dollar, since pricing has its own network effect and almost every stablecoin is already a dollar. The store of value pluralizes into whatever yields or holds best. And the medium of exchange, the thing you keep on hand to spend, thins toward a function that happens at settlement rather than an asset you own. Money becomes less a noun you hold than a verb that occurs.
The open question is which of the three becomes the center of gravity. Is the essence of money the ruler — the thing the world agrees to price in — or the thing held — the asset everyone stores wealth in? They have always been the same object. They may not stay that way.
Dalio’s objection: the gap that reappears
Here the argument has to stop and let its sharpest critic speak, because Dalio’s warning is not a footnote to this thesis — it is its limit.
The whole vision of just-in-time money rests on a hidden assumption: that conversion is always available at a fair price. Dalio’s point is that this is exactly what fails when it matters most. A bubble bursts not because the asset was worthless but because, all at once, everyone needs the spendable form and the market cannot manufacture it fast enough; the conversion that was frictionless in calm seizes in the panic. Paper wealth can rise far faster than the supply of money it implicitly claims to be worth, and the gap is invisible until the day everyone tries to cross it together.
That is the deepest case for the held medium of exchange, and frictionless conversion does not answer it — it may sharpen it. If everyone holds their wealth in assets and relies on instant conversion to spend, then the system’s true stock of spendable money shrinks to whatever can be converted right now, and a synchronized rush to convert is the very thing that breaks the rails it depends on. The medium of exchange, in this light, is not a tax we can finally stop paying. It is liquidity insurance we resent in every calm year and are grateful for in the one year it pays out.
So the question that the optimistic version glides past is the one worth holding onto: in a world where almost no one holds money, what is everyone’s money on the day they all need it at once — and who stands ready to provide it?
Where the power goes
Which returns the argument to the altitude of the first essay.
Monetary power has always lived in being the thing people are forced to hold. That is where seigniorage lives, where the float lives, where the leverage to decide who may pay whom lives. The dollar’s reach abroad and the bank’s spread at home are the same advantage in two settings: both rest on everyone needing to hold their money, in their form, on their rails.
If no one is forced to hold a medium of exchange, that power does not evaporate. It moves. It moves to whoever owns the unit of account — the standard the world quotes prices in — and to whoever owns the conversion layer — the rails, the solvers, the agents, the single wallet that Fink imagines holding everything you own. Note that this is not obviously decentralizing. The same letter that promises to democratize access also concentrates it: a world where one wallet holds all your assets is a world where whoever runs that wallet sits exactly where the bank and the correspondent network used to sit.
So the first essay’s question returns in a new form, and it is the one to end on rather than answer. Its subject was control over who may pay whom. This one’s is stranger and more fundamental: when holding money becomes optional, the old power attached to the held currency has to go somewhere. The contest of the next decade is over where — to the dollar as the world’s surviving unit of account, to the networks that convert, to the machines that decide the route, or to whoever owns the wallet that holds it all.
What is certain is only the change beneath the contest. For the first time in the history of money, the act that made us hold it — the costly, uncertain conversion of wealth into spendable form — is becoming nearly free. We have built our institutions, our crises, and our notion of money itself around the assumption that it never would be. That assumption is the thing now coming loose.